There is conventionally available an object detection device that acquires a thermal image for a detection range from a sensor device and detects a heat source, i.e., a target object, such as a human or the like from the thermal image. As a method of detecting an object using an object detection device, there is known a background difference method of previously recording a thermal image for a detection range at the time when no target object exists in the detection range, as a background image, and detecting a difference between the present thermal image acquired from a sensor device and the background image. The background difference method has an advantage in that it enjoys simple calculation and high detection accuracy under a stable environment.
In an object detection device that makes use of the background difference method, if the background image recorded earlier is continuously used as it stands, there is likelihood that erroneous detection may occur when the temperature within a detection range (hereinafter referred to as an “ambient temperature”) is changed by other causes than a target object. It is therefore necessary to update the background image.
In view of this, there has been proposed a device which has a function of updating a background image using an acquired image in case where a target object is not detected from the acquired image (see, e.g., Japanese Patent Application Publication No. 62-240823 (JP62-240823A)). Moreover, there has been proposed a device which has a function of acquiring a new image at a predetermined time after a previous background image is updated, calculating an arithmetic mean of the acquired image and the present background image and updating the background image to become a next background image (see, e.g., Japanese Patent Application Publication No. 10-308939 (JP10-308939A)). In the device disclosed in JP10-308939A, the step of calculating the arithmetic mean is not performed if a target object is detected from the acquired image.
In addition, as a device for identifying a human body from an image, there has been proposed a device which has a function of calculating a difference between a background image and a present acquired image, binarizing the difference with a threshold value and updating the background image with the present image with respect to the pixels which are equal to or smaller than the threshold value (see, e.g., Japanese Patent Application Publication No. 5-118916 (JP5-118916A)). The device disclosed in JP5-118916A further has a function of updating the background image with the present image with respect to the pixels which are not determined to be a human body even if the present image has the pixels which are larger than the threshold value (which are detected by the binarization).
In the configurations disclosed in JP62-240823A and JP10-308939A, however, the background image is not updated if the acquired image has a detected target object. For that reason, the update of the background image is not performed during the time when the target object continues to exist in a detection range. This makes it difficult to cope with a change in an ambient temperature. Thus, erroneous detection may possibly occur.
As in the device disclosed in JP5-118916A, even if the background image is updated with respect to the target object (human) non-detection region of an acquired image having a detected target object, the update of the background image is not performed with respect to the target object detection region of the acquired image. This makes it difficult to cope with a change in an ambient temperature. Thus, erroneous detection may possibly occur. For instance, if the ambient temperature goes down in a state that a human continues to exist in the same position, the body temperature of the human is also decreased along with the ambient temperature. However, the background image of the human-existing region is not updated. In this case, the difference between the human body temperature and the background image temperature is smaller than a threshold value. Thus, the object detection device may possibly make erroneous detection (issue an erroneous notice) that there exists no human. Conversely, if the ambient temperature goes up in a state that a human continues to exist in the same position, the temperature of the background image of the human-existing region becomes lower than the actual ambient temperature. Even if the human leaves the human-existing region, the object detection device may possibly make erroneous detection that the human still exists there.